Feature Selection and Feature Weighting Using Tunicate Swarm Genetic Optimization Algorithm With Deep Residual Networks

Feature selection (FS) method is applied for extracting only the relevant information from the dataset. FS seemed to be an optimization concept because appropriate feature selection is the significant role of any classification problem. Similarly, feature weighting is employed to enhance the classification performance along with FS process. In this paper, feature selection and feature weighting has been performed by integrated an optimization algorithm called tunicate swarm genetic algorithm (TSGA) with deep residual network (DRN). TSGA is the combination of tunicate swarm algorithm (TSA) and genetic algorithm (GA) incorporated to increase the performance of the classifier. This wrapper method-based feature selection and feature weighting techniques are performed to reduce the computation time as well as complexity. The effectiveness of the proposed method is estimated and compared with different methods such as TSA, CS-GA, and PSO-GA. The performance of DRN classifier is also validated and compared to existing classifiers like KNN, C4.5, and RF.